A platform for scientific application dissemination and community engagement

Such broad use has long been inhibited by tool complexity and by the fact that simulation and data analysis tools require complex pre- and post-processing and parallel workflow orchestration. Tedious tool installation procedures and complex parallel platform and data management requirements add further impediments to technology transfer and industrial adoption. Parallel Works Inc. has created a Web-based platform to encapsulate the entire process of using scientific and engineering modeling, simulation and analysis tools into shrink-wrapped software procedures. These automated procedures perform the higher-level workflow involved in using software applications, from installation to data preparation to procedure specification, execution, parallel iteration and optimization. They address and simplify analysis, visualization, collaboration, assessment, publication and dissemination. The proposed innovation research enables this platform to create for scientific applications what iOS and Android have achieved for mobile applications: a thriving marketplace where developers create and sell highly usable, broadly adopted high-value HPC solutions for industrial customers. The platform can be used from Web browsers, command line interfaces, and by APIs to service other application platforms. This Phase II R&D develops two key innovations to power the marketplace. A revolutionary, highly productive snap-together software “fabric” that enables scientific apps to be quickly and easily assembled into complete solutions that are far easier to use and adapt. And a framework of relocatable parallel computing workspaces that enables workflows with all their required components to be easily installed and executed at significant scales on elastic HPC computing resources from public providers and customer systems. The innovations integrate results from SciDAC and other DOE research to go far beyond what’s possible today. In Phase II, the focus will be on productization of a prototype fabric and workspace; continuing Phase I research on key problems of expressivity and interactivity (for productive scripting), and on achieving transparency between HPC and cloud resource providers. A large set of high-value energy, chemistry, and environmental applications will be engaged to obtain community requirements and feedback. Phase II research will implement the community engagement marketing roadmap evaluated in Phase I and lay the groundwork for numerous subsequent solution commercialization efforts. The research proposed in this SBIR has the potential to revolutionize and expand the practice of high-performance simulation and data analysis across multiple tiers of manufacturing and product design across the entire energy sector and environmental cleanup industry. It will create powerful ways to translate product and process design research into industrial productivity. The solutions derived from this research into workflow solution dissemination will drive efficiency in firms ranging from small engineering teams to large global manufacturing and energy enterprises, and to industries that extend beyond energy into automotive, aerospace, biomedical, pharmaceutical, financial and consumer domains, as well as to DOE labs and academic research.